作者：Cui, M., Zhu, M., Bi, S., Qin, B., Guan, Q. & Shi, K.

Monitoring total suspended solids (TSS) in lakes at a global scale is critical for understanding lake ecosystem responses to climate change and anthropogenic activities. However, reliable retrieval methods for TSS global mapping remain elusive due to the optical complexity of inland waters, leaving the spatiotemporal dynamics of global lake TSS poorly constrained. This study developed a global TSS retrieval model using a Random Forest (RF) algorithm based on Landsat OLI surface reflectance imagery. The model achieved satisfactory performance (R-2 = 0.73) and demonstrated exceptional robustness in estimating lake TSS concentrations (0-1,500 mg L--(1)) across diverse water quality conditions, geographical locations, and temporal ranges. During summers of 2014 to 2023, lakes in mid-low latitude regions, particularly arid zones, exhibited significantly higher mean TSS concentrations compared to other areas. Our analysis revealed that 62.9% of clear lakes (TSS < 10 mg L--(1)) that underwent statistically significant changes (P < 0.05) showed significantly increasing trends of TSS, emphasizing the urgent need to enhance monitoring and protection measures for clear-water ecosystems. Furthermore, we classified lakes into three types based on the predominant regulatory mechanisms controlling TSS dynamics and systematically elucidated the distinct mechanisms through which lake topography, hydrological conditions, climate, and watershed vegetation cover influence TSS concentrations across different lake types. This study provides new insights into the spatiotemporal patterns of global lake TSS variations and the response mechanisms of different lake types to hydrological conditions and climatic forcing, contributing improved understanding of global freshwater ecosystem dynamics under environmental change.

在全球范围内监测湖泊中的总悬浮固体（TSS）对于理解湖泊生态系统对气候变化和人类活动的响应至关重要。然而，由于内陆水体的光学复杂性，用于全球 TSS 制图的可靠反演方法仍难以实现，这使得全球湖泊 TSS 的时空动态受到很大限制。本研究基于Landsat OLI 地表反射率影像，采用随机森林（RF）算法开发了一个全球TSS反演模型。该模型表现出令人满意的性能（R²=0.73），并且在不同水质条件、地理位置和时间范围内估算湖泊TSS浓度（0-1500mg/L）时展现出极强的稳健性。2014至2023年的夏季期间，中低纬度地区的湖泊，尤其是干旱区的湖泊，其平均 TSS 浓度显著高于其他地区。我们的分析显示，在发生统计显著变化（P<0.05）的清澈湖泊（TSS<10mg/L）中，62.9% 的湖泊 TSS 呈现显著上升趋势，这凸显了加强清水生态系统监测和保护措施的迫切需求。此外，我们根据控制TSS动态的主要调控机制将湖泊分为三类，并系统阐明了湖泊地形、水文条件、气候和流域植被覆盖对不同类型湖泊TSS浓度影响的独特机制。本研究为全球湖泊TSS变化的时空模式以及不同类型湖泊对水文条件和气候强迫的响应机制提供了新的见解，有助于加深对环境变化下全球淡水生态系统动态的理解。

（来源：Water Research 2026   DOI: 10.1016/j.watres.2025.125274）